Plugin Development

[KevinH]

Hi,
Also make sure you are using Sigil 8.2 instead of 8.1. That error might be generated by using Sigil 0.8.1 but with the python launcher files from 0.8.2. So please verify you have installed Sigil0.8.2 properly and are running that version.

Kevin

[crutledge]

Originally Posted by Doitsu

This usually only happens if invalid .html files are processed with plugins.
1. Did you check the epub with epubcheck before you ran the plugin?
2. Did you install PunctuationSmarten_v0.1.0.zip?

If you checked the epub with epubcheck (not FlightCrew) and installed PunctuationSmarten_v0.1.0.zip you'll need to attach an epub that generates this error.

What is epubcheck and where do I find it?

[DiapDealer]

It definitely sounds like you're using some of the updated plugin launcher files with Sigil v0.8.1.

I just checked and the plugin will work with a stock version of Sigil 0.8.1 or a stock version of 0.8.2.

If you updated some of those launcher files (as bug fixes were coming out), you should probably just update to a stock version of Sigil 0.8.2.

The released version of the plugin worked OK for me on your attached epub using Sigil 0.8.2.

[Doitsu]

Originally Posted by crutledge

What is epubcheck and where do I find it?

It's the official ePub validator. It's available:

a) online

b) as a drag & drop Java app.

c) as a Sigil plugin

Note that thanks to the new Validation plugin feature in 0.8.2 the next version of the epubcheck plugin will behave like FlightCrew. I.e. you'll be able to click on an error message to display the line that the error refers to.

BTW, I also tested DiapDealer's plugin with the stock 0.8.1 and 0.8.2 versions and didn't have any problems.

[Toxaris]

I get more feedback on Linux than on Windows. The plugin kicks of a program and that program subsequently fires other programs. The output from those programs are hidden and not displayed anywhere. The program does not report that output back.
On Linux however I also get the output from those programs, but not on Windows.

[Doitsu]

@KevinH & user_none:

IMHO, Javascript would be the perfect embedded scripting language for Sigil, because it has been designed for website manipulation and the Qt Quick library already has JavaScript support.
Like Python it's cross-platform, relatively easy to learn, but unlike Python it doesn't require a separate HTML parser, which makes it perfect for HTML file manipulation.
Unlike the Python wrapper, which was designed to manipulate all parts of an epub, a hypothetical JavaScript wrapper would only need an HTML file iterator and the JavaScript equivalents of bk.readfile() and bk.writefile() to be useful.

Of course, since I'm not a programmer I have no idea how much work this would involve or whether this would be feasible at all.

Have you guys ever considered embedding a JavaScript interpreter?

[Toxaris]

Not being a Python programmer myself, I find it workable enough. Much better understandable than JavaScript anyway. The main irritation factor with Python is the indention however, but I think that just needs getting used too...

[Doitsu]

Originally Posted by Toxaris

Not being a Python programmer myself, I find it workable enough.

I really like Python, but if you want to do some serious HTML parsing, you'll need to use external libraries such as Beautiful Soup. The older version, BS3, can be embedded, but the current version, BS4 is no longer embeddable.
You can do pretty much all that BS does with JavaScript without external libraries. (Python also has DOM manipulation libraries but unlike BS they can't handle Unicode.)

Originally Posted by Toxaris

The main irritation factor with Python is the indention however, but I think that just needs getting used too...

Initially, I found it annoying, too, but you get used to it and it forces you to write structured code.
There are other things that I find much more illogical. For example, string slicing parameters:

word = 'Python' 
print word[0:2]
# prints Py not Pyt

[Toxaris]

I am used to using additional libraries though... For HTML handling I use HTML Agility Pack, which is quite good but has some quirks. It was always my intent to learn more about Python, but I didn't have a good excuse. Of course I only have done baby steps so far.

[KevinH]

Hi,

Some points:

- Even Javascript requires a dom parser of some sort. Since javascript is typically embedded inside web pages, this is typically provided for it, not inherent. Something must create the dom it is wrapped around.

- BS4 is embeddable quite easily and will probably become part of the python 3 future of Sigil. Also embeddable quite easily is html5lib. Under python 3 built-in dom and xpath routines are quite powerful and fully unicode aware.

- most dom implementations like etree can easily be built from any tokenizer/parser including the built in quickparser.py support or html5lib or ... This is in fact what both BS3 and BS4 does.

So the real reason python was chosen over other languages for the first interface was that it is simple to learn, extremely powerful, easy to extend, and that has a ready flexible set of xml/xhtml processing libraries built in or easily addable.

Hope this helps,

KevinH

[Doitsu]

Originally Posted by KevinH

- Even Javascript requires a dom parser of some sort. Since javascript is typically embedded inside web pages, this is typically provided for it, not inherent. Something must create the dom it is wrapped around.

I didn't know this. Thanks for your detailed explanation.

Originally Posted by KevinH

- BS4 is embeddable quite easily and will probably become part of the python 3 future of Sigil. Also embeddable quite easily is html5lib. Under python 3 built-in dom and xpath routines are quite powerful and fully unicode aware.

BS3 is easily embeddable in a plugin, because it's a single Python file, but BS4 installs a folder. Is there any Python tool that'll "flatten" that folder into a single Python file or will I need to change the Import statement after copying the complete BS4 folder into the folder of the plugin?
(The default Import statement didn't work when I copied the whole BS4 folder into the plugin folder.)

[KevinH]

Originally Posted by Doitsu

BS3 is easily embeddable in a plugin, because it's a single Python file, but BS4 installs a folder. Is there any Python tool that'll "flatten" that folder into a single Python file or will I need to change the Import statement after copying the complete BS4 folder into the folder of the plugin?
(The default Import statement didn't work when I copied the whole BS4 folder into the plugin folder.)

There are a number of ways to handle it. You can manually add the subdirectory to sys.path so that a normal import will find it or alternatively you can use module based relative paths for importing it. The same is true for libgui as well.

I have added two different ways to import from a subdirectory either as a folder or as a module. See libgui_as_folder_example.zip and libgui_as_module_example.zip.

Notice the differences in test.py for how you add to the path or do the module based import. Also look at the differences in libgui as module that has its own __init__.py and uses relative addressing to import from .askfolder and .user_utilities, versus the libgui as folder example with no __init__.py and that relies on the sys.path search being done to find its imports.

Take care,

KevinH

ps. please note that for BS4 to work crossplatform as pure python, it is probably a good idea to include html5lib as well to get a pure python parser/tokenizer to inside BS4. Together, these will work on all platforms, and make embedding of BS4 quite easy.

[Doitsu]

Originally Posted by KevinH

I have added two different ways to import from a subdirectory either as a folder or as a module. See libgui_as_folder_example.zip and libgui_as_module_example.zip.

Thanks for your examples. I've finally figured out how to embed BS4.

[turbulent]

In an output sigil plugin, is it possible to get the file name of the epub file currently opened by sigil? So it's convinient to name the generated file after the epub file.

[Toxaris]

Originally Posted by turbulent

In an output sigil plugin, is it possible to get the file name of the epub file currently opened by sigil? So it's convinient to name the generated file after the epub file.

Not that I can recall. It would be of limited use anyway, since it cannot be used as name to save the output. Reason is simple, as the file is still open within Sigil.